The effect of flow on swimming bacteria controls the initial colonization of curved surfaces

Abstract

The colonization of surfaces by bacteria is a widespread phenomenon with consequences on environmental processes and human health. While much is known about the molecular mechanisms of surface colonization, the influence of the physical environment remains poorly understood. Here we show that the colonization of non-planar surfaces by motile bacteria is largely controlled by flow. Using microfluidic experiments with Pseudomonas aeruginosa and Escherichia coli, we demonstrate that the velocity gradients created by a curved surface drive preferential attachment to specific regions of the collecting surface, namely the leeward side of cylinders and immediately downstream of apexes on corrugated surfaces, in stark contrast to where nonmotile cells attach. Attachment location and rate depend on the local hydrodynamics and, as revealed by a mathematical model benchmarked on the observations, on cell morphology and swimming traits. These results highlight the importance of flow on the magnitude and location of bacterial colonization of surfaces.

Fig. 1. Microfluidic model of bacterial colonization on curved surfaces in the presence of fluid flow.

a Schematic of the microchannel containing pillars of different diameters, d_P (200 µm, 100 µm, and 50 µm; each repeated 2 times). The order of pillars with respect to the flow direction has shown no influence on the observed phenomenon. Flow velocity, U (b) and radial shear rate, S_R (c) around a 100-µm pillar, computed with COMSOL Multiphysics at a mean flow velocity of 500 µm s⁻¹. Superimposed arrows indicate the local velocity field. d Fluorescent image of GFP-tagged P. aeruginosa PA14 wt cells attached to a 100-µm pillar after 5 h flow at rescaled velocity U/V = 3.3 (U = 150 µm s⁻¹).

Fig. 2. Surface colonization on pillars by P. aeruginosa is determined by bacterial motility.

Intensity distribution of the fluorescent signal from GFP-tagged motile (PA14 wt, a) and nonmotile (PA14 flgE, b; PA14 motB, c) P. aeruginosa cells attached to a 100-µm pillar after 5 h flow at a rescaled flow velocity of U/V = 6.6. d, e Angular distribution of the normalized attachment density of bacteria on the pillar obtained with a mathematical model for motile (d) and nonmotile (e) cells for the same flow rate and pillar dimension as a–c. f Integrated intensity, I_IN obtained for the motile (wt) and nonmotile (flgE, motB) strains from the images in a–c. Error bars correspond to the standard error of the mean.

Fig. 3. Capture efficiency of P. aeruginosa depends on pillar dimension and flow velocity.

a Experimental capture efficiency, η_C^exp, of motile P. aeruginosa PA14 wt cells, as a function of the rescaled flow velocity U/V, for pillars of diameter 50 µm (blue circles), 100 µm (red squares) and 200 µm (black triangles), measured after 5 h from the start of the experiment. The dashed curve shows the scaling ${\eta }_{\mathrm{C}}^{\exp }~{\left(U/V\right)}^{-1}$. b Capture efficiency, η_C^mod, as a function of U/V, obtained from the model for the same pillar diameters as in (a). c Experimental capture efficiency, η_C^exp, as a function of the Péclet number, Pe, for motile PA14 wt (filled symbols) and nonmotile (PA14 flgE, open symbols with a cross; PA14 motB, open symbols) cells, for pillars of different diameters. d Capture efficiency, η_C^mod, as a function of the Pe, obtained from the model for different pillar diameters, in the case of motile (filled symbols) and nonmotile (open symbols) cells. Vertical dotted lines in c and d represent the Pe numbers corresponding to U/V = 20, calculated for each pillar dimension. Error bars correspond to the standard error of the mean.

Fig. 4. Fluid shear affects the trajectories of swimming bacteria around a pillar.

a Sample trajectories and orientation of P. aeruginosa PA14 wt cells in flow around a 100-µm pillar at U/V = 3.3 (U = 150 µm s⁻¹), for motile (warm colors) and nonmotile (blue) cells. b Trajectories of motile (black) and nonmotile (gray) cells, simulated without rotational noise ξ_R, in flow around a 100-µm pillar at U/V = 3.3 simulated with the model. The color scale represents the radial shear rate, S_R, around a 100-µm pillar at U = 500 µm s⁻¹, reported in Fig. 1c.

Fig. 5. Fluid velocity modifies the angular distribution of bacterial colonization around a pillar.

Angular distribution of the fluorescence intensity, I (blue; experiments after 5 h of flow of a diluted suspension of PA14 wt GFP cells) and the simulated attachment density (orange; model) on a 100-µm pillar for a mean rescaled flow velocity U/V of 3.3 (a), 6.6 (b), 13.3 (c) and 22.2 (d). e Normalized standard deviation σ_θ/σ_u for motile PA14 wt bacteria as a function of the rescaled flow velocity, U/V, for pillars of diameter 50 µm (blue circles), 100 µm (red squares) and 200 µm (black triangles).  (Insets) Intensity distribution I of PA14 wt cells, normalized by the maximum value (I_max), attached to a 100-µm pillar at U/V = 6.6 (f) and U/V = 22.2 (g). At moderate flows, preferential colonization occurs on the leeward side of the pillar. Flow direction is from left to right. h Normalized standard deviation σ_θ/σ_u obtained with the model as a function of U/V and for the same pillar dimensions as in e. Results for motile cells (filled symbols) and for nonmotile cells (open symbols) are shown. (Inset) σ_θ/σ_u of the experimental intensity angular distribution as a function of σ_θ/σ_u from the model for different pillar diameters. The dashed line represents y = x. Error bars correspond to the standard error of the mean.

Fig. 6. Surface properties of the pillar do not affect the angular distribution of bacterial colonization.

a Behavior of tryptone broth droplets on PDMS surfaces measured 5 days (upper panel; contact angle 95° ± 5°) and 1 h (lower panel; contact angle 15° ± 5°) after plasma treatment showing, respectively, the hydrophobic and hydrophilic nature of the two surfaces. In the lower panel, the droplet wets the surface creating a film: its borders are marked with a blue dashed line in the image for ease of visualization. b Angular distribution of the fluorescence intensity, I, of PA14 wt GFP cells attached on a hydrophobic pillar (blue) and on a hydrophilic pillar (orange), for a flow velocity at U/V = 6.6 and a pillar diameter of 100 µm. c Surface coverage (measured on the upper surface of the microfluidic channel) for the hydrophobic surface (blue) and the hydrophilic surface (orange) under the same experimental conditions as b. d Slabs of PDMS containing 3% curing agent (upper panel; Young modulus = 150 ± 50 kPa) and 10% curing agent (lower panel; Young modulus = 2.25 ± 0.25 MPa) undergoing compression from a binder clip in order to visualize the difference in stiffness. e Angular distribution of the fluorescence intensity, I, of PA14 wt GFP cells attached on a pillar containing 3% (yellow), 5% (green) and 10% (blue) curing agent, for a flow velocity at U/V = 6.6 and a pillar diameter of 100 µm. f Surface coverage (measured on the upper surface of the microfluidic channel) for PDMS containing different concentrations of curing agent under the same experimental conditions as e. Angular distributions and surface coverages were measured after 5 h of continuous flow. Error bars correspond to the standard error of the mean.

Fig. 7. Preferential leeward attachment on a sinusoidal surface by motile E. coli.

a Schematic of the microchannel with a sinusoidal lateral surface with wavelength 50 µm and amplitude of 25 µm. b Distribution of the fluorescence intensity from GFP-tagged E. coli wt cells attached to the sinusoidal surface after 3 h of flow at U = 150 µm s⁻¹ (U/V = 6.9), averaged over 100 identical periods and normalized for the maximum intensity value. c Normalized attachment density of cells on the sinusoidal surface obtained from the model at U = 150 µm s⁻¹. d Radial shear rate, S_R, around one period of the sinusoidal lateral wall, computed with a finite element method at U = 500 µm s⁻¹. Superimposed arrows indicate the local velocity field. e Simulated trajectories of spherical nonmotile (blue) and of elongated motile (yellow) cells in flow around a period of the sinusoidal wall. f Simulated attachment density of spherical nonmotile and motile bacteria on the sinusoidal wall at U = 150 µm s⁻¹.

Fig. 8. Colonization of a randomly corrugated surface by E. coli.

a Fluorescent image acquired at channel mid-plane of GFP-tagged E. coli wt cells attached to the lateral corrugated surface of a microfluidic channel (height = 100 µm) after 3 h of flow of a diluted bacterial suspension at a mean flow velocity of U = 150 µm s⁻¹. b Intensity distribution of the fluorescent signal shown in a, normalized by its maximum value. c Normalized attachment density of cells (elongation q = 8.5, swimming speed V = 21.6 μm s⁻¹) on the corrugated surface obtained with the model at a mean flow velocity of U = 150 µm s⁻¹ (U/V = 6.9).